Investigation of Deep Learning Techniques  Used in Medicinal Plants Identification

The identification and classification of medicinal plants play a significant role in the field of healthcare and traditional medicine. This study investigates the use of deep learning techniques for the identification and classification of medicinal plants, utilizing various pre-trained models to achieve high accuracy in classifying plant species. The implemented algorithms include VGG16_ResNet50, VGG16_MobileNetV2, VGG16_LSTM, and VGG16_EfficientNetB0, each designed to leverage the strength of convolutional neural networks (CNNs) and transfer learning for feature extraction and classification tasks. The dataset comprises 16 medicinal plant species, including 'Arive-Dantu', 'Basale', 'Betel', 'Crape_Jasmine', 'Curry', 'Mint', 'Neem', 'Oleander', 'Parijata', 'Peepal', 'Pomegranate', 'Rasna', 'Rose_apple', 'Roxburgh_fig', 'Sandalwood', and 'Tulsi', which are commonly used in traditional medicine.

The VGG16-based models with different architectures were fine-tuned to enhance classification performance, and the LSTM model was employed to capture sequential patterns in plant features, further improving classification accuracy. The study also demonstrates the robustness of these models in distinguishing between plant species with similar features. The proposed deep learning-based solution offers a promising approach to automate the process of medicinal plant identification, making it scalable and efficient for real-world applications in healthcare, agriculture, and biodiversity conservation.

Keywords: Medicinal Plant Identification, Deep Learning, VGG16, ResNet50, MobileNetV2, LSTM, EfficientNetB0, Classification, Transfer Learning, Feature Extraction, Sequential Learning, Plant Species Classification, Healthcare, Biodiversity Conservation.